隨著智慧型手機問世及可攜式電子產品快速增加,電子零組件的世代更迭速度也加快了。由於終端產品的生命週期縮短,導致取得訂單的因素除了品質外,更要能快速出貨。因此,在排除生產缺陷問題的時間被大幅壓縮。意指需要在有限的時間內找出問題並排除問題。 因此本研究著重在如何使用資料探勘,分析目前生產發生的缺陷問題以及如何藉由過去產品缺陷資料找出目前生產發生缺陷問題的解決方法。本研究個案為印刷電路板廠,利用案例式推論架構,以決策樹方法分析過去生產有缺陷問題的產品與生產機台的關係,推導出可能發生問題的站點及機台組合,再利用K-means分群法建立過去缺陷案件的索引。藉由過去發生過相似的缺陷,以及過去解決缺陷的經驗,用來解決目前的缺陷及探討與缺陷發生相關的生產條件,進而達到快速查詢缺陷案件的處理方式,最終獲得客戶滿意度提昇、訂單量增加、節省處理異常的時間及訓練人力的成本。
With the advent of smart phones and the rapid increase in portable electronic products, the generational change of electronic components has also accelerated. Due to the shortening of the life cycle of the terminal product, the factors that lead to obtaining orders are not only the quality, but also the ability to ship quickly. As a result, the time required to eliminate production defects has been greatly reduced. It means that the problem needs to be found and eliminated within a limited time. Therefore, this research focuses on how to use data exploration, analyze current production defects and how to use past product defect data to find solutions to current production defects. This research case is a printed circuit board factory, using a case-style inference structure, using a decision tree method to analyze the relationship between the defective products and the production machine in the past, deduce the site and machine combination that may have problems, and then use K -Means grouping method establishes an index of past defect cases. With similar defects in the past and experience in solving defects in the past, it is used to solve current defects and discuss production conditions related to the occurrence of defects, and then achieve a quick way to query defect cases, and finally obtain customer satisfaction and order Increase the volume, save the time for handling exceptions and the cost of training manpower